Next-generation sequencing (NGS) has failed to detect mesenchymal epithelial transition factor gene (MET) polysomy in previous studies. We included three non-small cell lung cancer (NSCLC) cohorts in this retrospective study to establish new criteria for detecting MET polysomy and to explore the clinical relevance of MET polysomy. Cohort 1 included 53 patients whose tissues were available for both fluorescence in-situ hybridization (FISH) and NGS assays. Paired plasma and tissue samples were obtained from 261 patients with NSCLC as cohort 2. Cohort 3 included 46 patients with metastatic NSCLC, who presented with MET copy number gain assessed by NGS. Receiver operating characteristic (ROC) analysis demonstrated that a cut-off point of 2.3 copies achieved the maximum Youden index in discriminating polysomy from normal copy number. Compared with the FISH test for MET polysomy, the sensitivity, specificity, and agreement of NGS were 90%, 90%, and 96.2%, respectively. Following optimization using maximum somatic-allele-frequency (MSAF), the sensitivity and specificity of NGS for defining polysomy using plasma samples according to different ctDNA mutation frequencies were 42% and 63%. The concordance rate between tissue and plasma samples for detecting polysomy was 85%. Regarding the response to MET inhibitor, the median progression-free survival (PFS) of the MET amplification group was significantly higher than that of the polysomy group. The median PFS was similar between the polysomy and normal groups. Our results indicated that NGS may serve as an alternative method for detecting MET polysomy in NSCLC tissues. Moreover, patients with MET polysomy may not benefit from MET inhibitors.